Your search bar leaks revenue when a shopper searches, finds nothing useful and leaves. You can size that loss with data you already have: how many searches return no results, how many shoppers who search go on to buy, and what an average order is worth. This article turns those numbers into a rough yearly figure for your site search revenue.
I work on this problem for e-commerce teams through my search and discovery consulting. The method below is deliberately simple. The first job is to find out whether the leak is big enough to be worth fixing, not to build a perfect model.
The short version: count the failed searches, estimate what a successful search is worth, and multiply. The result is an estimate to argue with, not an exact measurement, but it is enough to decide whether search deserves your next quarter.
What a failed search actually costs
Shoppers who use site search often have strong purchase intent. They have already decided what they want and are asking your site for it. When search fails them, they do not file a support ticket. They try another phrase, browse by category, or leave. Only the first two keep the sale.
Baymard Institute, which tests e-commerce sites with real users, found that nearly 50% of sites fail to provide users with effective ways to recover from a search that yields no results. It notes that poorly handled no-results pages often lead to site abandonment. So the cost of a failed search is the sale you lose when the shopper leaves, plus the extra work you put on the sales you keep.
Where searches fail
Failed searches come in three kinds, and each needs a different number to find.
No results.
The search returns an empty page. This is the easiest failure to count, because any search log can tell you which queries came back empty.
Wrong results.
Results appear, but the product the shopper meant is far down the list or missing. Nothing flags this as a failure, so you find it through behavior: the shopper searches again straight away, or leaves the results page.
Slow results.
Results arrive late enough that the shopper retypes or gives up. This one shows up as response time rather than in the queries themselves.
This article sizes the first two. Fixing them is a separate job, and I cover it in the e-commerce search health check, so I will not repeat that checklist here.
How to size the leak with your own data
You need four numbers for one month of traffic. Pull them from your search logs and analytics.
1. Searches
The total number of searches in the month. In Google Analytics 4, enhanced measurement can collect site search as a view_search_results event with a search_term parameter. It triggers on query parameters such as q, s, search, query or keyword in the URL.
2. Zero-result searches
How many of those searches returned nothing. Analytics that only see the URL cannot tell a results page from an empty one, so use your search tool's own log, or ask your developers to send an event from the no-results page.
3. Search-to-purchase rate
The share of sessions with a search that end in a purchase. Use the same month, and compare it with sessions that did not use search.
4. Average order value
What an order is worth, taken from the same sessions.
A worked example
Multiply the zero-result searches by the purchase rate of a search that does return results, then by the average order value. That gives the most the leak could cost. Then decide how much of it you could realistically recover. Here is the arithmetic with made-up numbers.
Searches per month
200,000
Zero-result searches (8%)
16,000
Purchase rate for a search with results
4%
Average order value
$80
Ceiling: 16,000 × 4% × $80
$51,200 a month
If you recover a quarter of it
$12,800 a month, about $154,000 a year
Every number above is illustrative, not a client result. Replace them with yours. If the answer is small, search is not your biggest leak and you have saved yourself a project. If it is large, you now have a figure to put in front of whoever owns the budget. One caution: the same shopper retrying a query counts several times, so count sessions or unique searchers if your data allows.
Sizing wrong results
Wrong results do not show up as an empty page, so count reformulations instead: searches followed by another search in the same session. A high share means shoppers are not getting what they wanted the first time. Treat it as a warning until you compare the purchase rate of sessions with a reformulation against sessions without one.
Which fixes to try first
Rank fixes by how much of your measured loss each one addresses, not by which is easiest. If most failed searches are misspellings, typo tolerance comes first. If they are for products you do not stock, merchandising and demand analysis do. If they use words your catalog does not, synonyms do.
The search health check walks through each of these. If you run an e-commerce or marketplace site and would rather have the structured version, see my work for e-commerce and marketplace teams, or score your own site with the e-commerce search audit.
What to measure after the fix
Run the same four numbers again on a month of the same length, and compare the zero-result rate, the search-to-purchase rate and the share of reformulations. Change one thing at a time where you can, so you know which fix moved the number. Seasonality can move these figures by itself, so compare the same period of the year, or split traffic if you have enough of it.
Common questions about search revenue loss
How do I tell if my search is losing sales?
Pull a month of searches, the zero-result searches among them, and the purchase rate and order value for sessions that used search. Multiply them as in the worked example. If the result is a meaningful share of revenue, search is leaking.
What zero-result rate is a problem?
I do not have a threshold I can source, so I would not borrow one. A rate that looks small can still be an expensive leak on high traffic, so size it in money using the method above.
Which fix should I start with?
The one that addresses the largest share of your failed searches. Group the failed queries by cause, such as misspellings, missing products and catalog wording, and start with the biggest group.
Do I need Google Analytics to do this?
No. Your search tool's own query log is enough for the first two numbers. Google Analytics 4 can record site search through enhanced measurement, but it cannot see an empty results page unless you send it an event.
